Biogeography and variability of eleven mineral elements in plant

Ecology Letters, (2011) 14: 788–796
doi: 10.1111/j.1461-0248.2011.01641.x
LETTER
Biogeography and variability of eleven mineral elements in plant
leaves across gradients of climate, soil and plant functional type
in China
W. X. Han,1,2 J. Y. Fang,1,*
P. B. Reich,3,4 F. Ian Woodward5
and Z. H. Wang1
Abstract
Understanding variation of plant nutrients is largely limited to nitrogen and to a lesser extent phosphorus. Here
we analyse patterns of variation in 11 elements (nitrogen ⁄ phosphorus ⁄ potassium ⁄ calcium ⁄ magnesium ⁄
sulphur ⁄ silicon ⁄ iron ⁄ sodium ⁄ manganese ⁄ aluminium) in leaves of 1900 plant species across China. The
concentrations of these elements show significant latitudinal and longitudinal trends, driven by significant
influences of climate, soil and plant functional type. Precipitation explains more variation than temperature for
all elements except phosphorus and aluminium, and the 11 elements differentiate in relation to climate, soil and
functional type. Variability (assessed as the coefficient of variation) and environmental sensitivity (slope of
responses to environmental gradients) are lowest for elements that are required in the highest concentrations,
most abundant and most often limiting in nature (the Stability of Limiting Elements Hypothesis). Our findings
can help initiate a more holistic approach to ecological plant nutrition and lay the groundwork for the eventual
development of multiple element biogeochemical models.
Keywords
Biogeochemistry, biogeography of plant chemistry, climate, plant nutrients, plant functional type, soil nutrient
availability, soil pH, stability of limiting elements, stoichiometry, variability in leaf mineral elements.
Ecology Letters (2011) 14: 788–796
Mineral nutrients sustain all plant life on our planet (Aerts & Chapin
2000; Epstein & Bloom 2004). It is necessary to maintain sufficient
concentrations and relatively stable nutrient ratios in plant tissues
(stoichiometric balance) for healthy growth (Marschner 1995).
However, taxa may differ in the need for, and capability of obtaining
and maintaining, specific ranges of concentrations and ratios of
different nutrient elements in the plant body (stoichiometric homeostasis) (Sterner & Elser 2002). A better understanding of variation of
all essential plant nutrients is critical to the development of a broad
(rather than nitrogen-centric) perspective on the ecology of plant
nutrition, as well as in long-range, holistic biogeochemical models.
However, such understanding to date is largely limited to nitrogen (N)
and to some extent phosphorus (P).
As all essential nutrient elements play a role in plant health and
ecosystem function, a nuanced framework for understanding current
biogeochemical cycles, including carbon (C) and N, and how they will
change in the future, requires improved understanding of these
elements and their interactions. Although globally N and P are
considered of paramount importance to plant function, it is widely
known that many other elements are also important in specific
contexts or regions, either due to limitations or toxicity, or impacts on
C ⁄ N ⁄ P cycling (Lynch & St. Clair 2004; Vitousek et al. 2010). For
example, tissue calcium (Ca) and magnesium (Mg) deficiency and
manganese (Mn) and aluminium (Al) toxicity are common in certain
highly leached tropical soils (Lynch & St. Clair 2004); tissue Ca can act
as a regulator of soil pH and cation exchange capacity (Reich et al.
2005); tissue molybdenum (Mo) and iron (Fe) can influence N fixation
response to rising CO2 (Hungate et al. 2004; Van Groenigen et al.
2006; Barron et al. 2009), and micronutrients added to a tropical forest
enhance leaf litter decomposition and leaf nitrogen content (Kaspari
et al. 2008). Thus it is imperative that we begin to focus our attention
in the direction of the full set of mineral elements (Lynch & St. Clair
2004; Ågren 2008; Townsend et al. 2011).
Despite the examples given above of the importance of elements
beyond C and N, their broad patterns of tissue concentration and
stoichiometry are very poorly documented compared with C and N
(McGroddy et al. 2004; Reich 2005; Elser et al. 2007). Broadly
speaking, it is not well understood how or why the biogeography
(including both means and variation) of different plant minerals is
created and maintained, nor whether patterns should differ for
different elements (Lynch & St. Clair 2004; Marschner & Rengel
2007).
China spans large gradients of climate and vegetation, from tropical
rainforest to cold alpine meadow or dry Gobi desert, covering nearly
1
4
INTRODUCTION
Key Laboratory for Earth Surface Processes, Ministry of Education, Department
Hawkesbury Institute for the Environment, University of Western Sydney,
of Ecology, Peking University, Beijing 100871, China
Richmond, NSW 2753, Australia
2
5
Key Laboratory of Plant-Soil Interactions, Ministry of Education, Key Labora-
Department Animal & Plant Sciences, University of Sheffield, Sheffield,
tory of Plant Nutrition, Ministry of Agriculture, College of Resources and
S10 2TN, UK
Environmental Sciences, China Agricultural University, Beijing 100193, China
*Correspondence: E-mail: [email protected]
3
Department of Forest Resources and Institute on the Environment, University
of Minnesota, Minnesota 55108, USA
! 2011 Blackwell Publishing Ltd/CNRS
Biogeography and variability of leaf chemistry 789
Letter
all the vegetation types in the world (Fang et al. 2002; Zhang 2007).
It thus provides a good representation of much of global biome
heterogeneity and a unique opportunity to examine how climate, soil
and plant species interact in controlling leaf chemistry. The variation
results from north-south and east-west gradients in climate, as well as
variation in geomorphology and soil substrate materials, and resulting
plant compositional variation (Hou 1983). Over a prolonged period, a
large body of plant nutrient data has been accumulated in China.
These data were obtained for leaves of nearly 2000 plant species
across China (Fig. S1), coupled with information on location, climate
and soil nutrients, involving concentrations of 11 mineral elements
[N, P, potassium (K), Mg, Ca, sulphur (S), silicon (Si), Fe, sodium
(Na), Mn and Al].
Herein, we first explore the biogeographic patterns of multiple
elements in plants at the national scale; and whether and how these
elements show similar or differential heterogeneity among plant
functional groups and sensitivity to environmental factors (e.g. climate
and soil nutrient availability). We then investigate variations in these
elements and their possible differential responses to the environmental factors. We hypothesise that nutrients required in higher
concentrations in leaves and considered most frequently limiting in
nature should show smaller variations in their concentration and lower
sensitivity along environmental gradients than the elements at the
opposite end of the spectra (Stability of Limiting Elements Hypothesis). We examine the hypothesis by analysing the variation in the
11 plant minerals and their responses to the gradients of climate and
soils.
MATERIAL AND METHODS
Concentrations of 11 plant leaf minerals
The concentrations of 11 leaf minerals (N, P, K, Ca, Mg, S, Si, Fe, Na,
Mn and Al) in 1900 higher terrestrial plant species, belonging to 788
genera and 175 families, at 752 sites across China, were obtained from
our field measurements and published literature (Fig. S1; see also
Data S1 for details). In total, 4796 records on the whole, or about
2392 observations for each mineral on average, are involved in the
dataset. The leaves for mineral analyses in this dataset were sampled
during the growing season (June to September, mostly July and
August). Site-related information, including the longitudes, latitudes,
climate and soil mineral background values, were also documented in
the dataset.
Climatic variables, soil data and functional types
Geographic patterns of leaf minerals may be related to climatic
variables, including temperature, precipitation, length of growth
season and climatic variability (Reich & Oleksyn 2004). In this study,
five climatic variables were employed to analyse the climatic controls
on leaf mineral spatial patterns: mean annual temperature (MAT, "C),
mean annual precipitation (MAP, mm), growing season length (GSL,
days), and average diurnal range of temperature (DRT, "C) and annual
precipitation seasonality (coefficient of variation of monthly mean
precipitations) (APS, %).
For sites where MAT ⁄ MAP and latitude ⁄ longitude were recorded,
these values were used for the analyses. For the records lacking
detailed geographic coordinates, we used the latitude ⁄ longitude of
the geographical centre of the sample areas. For the sites where
MAT ⁄ MAP were not recorded, estimates of MAT ⁄ MAP were
extracted from a global climate dataset with a resolution of
0.0083 · 0.0083 (ca. 1 km · 1 km) obtained from http://www.
worldclim.org/. GSL was defined as the number of days with diurnal
mean temperature > 5 "C, and together with DRT and APS, was
estimated with records of 740 climatic stations in China (during 1950–
1999) using a Kriging extrapolation method.
Soil N, P and K data were obtained from the national soil survey
and our field measurements; other soil minerals were from another
national soil survey, except that soil S data cover only part of the
country and were collected from several separate studies (for
details, see Fig. S2 and the corresponding Supplementary references).
All species in the dataset were primarily classified into seed plants
and ferns; seed plants were further divided into six groups according
to their respective functional types. The four woody plant groups are
deciduous broadleaves, deciduous conifers, evergreen broadleaves and
evergreen conifers; the herbaceous groups are grasses (families of
Cyperaceae and Gramineae) and forbs (all others).
Indices of physiological requirement and relative limitation
Both physiological (Ingestad 1997) and ecological (Sterner & Elser
2002) stoichiometry should increasingly constrain variability as
elemental relative supply limitations grow and requirements increase.
To explore such a trend, we used an index of physiological
concentration requirement (following Marschner 1995; also see
supplementary Appendix S1). We use this without implying that the
absolute values hold for all taxa and all conditions, but instead
proposing that the index is useful because it provides a relative
measure of general physiological requirement. We also used a ranking
of the elements (index of relative limitation) from those considered
most (1) to least (8) frequently limiting in terrestrial plants
(N > P > K > Ca > Mg > S > Fe > Mn). This relative limitation
order differs from the rank order by requirement only in that P is
moved ahead of K and Ca. We developed this ranking based on our
synthesis of information from various sources (e.g. Marschner 1995;
Jobbágy & Jackson 2001; White & Brown 2010; Townsend et al.
2011). Because this ranking is somewhat subjective, we developed two
other rankings for comparison: (1) searching the number of fertiliser
studies by individual elements as well as the relative amount of
fertiliser consumption, and (2) using total soil contents of these
elements as an indicator of soil supply potential (see supplementary
Appendix S1 for details). Both the indices (of physiological
requirement and of relative limitation) are limited to the eight
elements known to have specific physiological requirements
and whose limitations in soils are best understood. Results shown
herein are similar if rank order of some elements is reversed, so the
conclusions are not dependent on either choice of or the absolute
accuracy of the rank order. The coefficient of variation (CV) was
calculated for each element as the metric for the variation in chemical
concentration.
Data analysis
All leaf mineral concentrations were log10-transformed before analyses
to improve the data normality. The leaf mineral concentrations were
averaged at the species or species-by-site (site-species) level in the
same way as Han et al. (2005). As carbon concentration is relatively
! 2011 Blackwell Publishing Ltd/CNRS
790 W. X. Han et al.
Letter
stable, we use elemental concentrations as our index of stoichiometry
relative to carbon.
Stepwise multiple regressions were applied to identify the most
influential climatic variables among the five climate variables (MAP,
MAT, GSL, APS and DRT). To explore the possible effects of soils
on leaf minerals, Spearman!s rank correlations were performed
between leaf minerals and the corresponding soil mineral background
contents at the national scale; Student–Newman–Keuls (S–N–K)
post hoc tests were then employed to compare the leaf mineral
concentrations among plants growing in soils with different soil
mineral levels for each single element. To demonstrate the relative
effects of climate, soil and species composition (functional type),
partial general linear model (GLM) analyses were applied. Partial
GLM separates the variance explained by different factors into the
independent effects of each individual factor and interactive effects
between factors.
Considering that global scale modelling and understanding of
vegetation function is confined to functional type groups (e.g.
Woodward 1987), it is important to analyse the data at this level of
classification and given the study goals it is also a useful attempt to
assess the indirect effect of climate on plant nutrient status through
climate impacts on the distribution of functional types. For such
purposes, we analysed how leaf chemistry responds to climate and soil
chemistry by functional type. Because the reduced major axis (RMA)
regression slopes of the relationship between leaf chemical concentration and temperature, precipitation and soil chemistry can be used
to indicate the response of leaf chemistry to variation in climate and
soil chemistry (Sokal & Rohlf 1995), we calculated these slopes for all
the 11 elements for five functional types (deciduous broadleaf,
evergreen broadleaf, evergreen conifer, grass, and forb) but not for
deciduous conifers and ferns due to their small sample sizes. Positive
RMA slopes at the functional-type level indicate increases in chemical
concentration with increasing temperature, precipitation or soil
elemental content level, and vice versa. RMA slopes for the three
variables (temperature, precipitation, and soil) were transformed to
eliminate effects of the different unit of MAT, MAP and soil nutrient
level (For full details, see supplementary Appendix S1).
All analyses were conducted with statistical software SPSS 13.0 (SPSS
Inc., Chicago, IL, USA, 2004) and R 2.2.1 (R Development Core
Team, 2005). For full details, supplementary Appendix S1.
RESULTS
Statistics and biogeographic patterns of leaf minerals
The mean concentrations of the 11 leaf minerals in China!s plants vary
greatly – from 0.11 mg g)1 (or 0.002 mol kg)1) for Mn to
20.5 mg g)1 (or 1.463 mol kg)1) for N (Fig. 1 and Table S1), with
ratios of N : P : K : Ca : Mg : S : Si : Fe : Na : Mn : Al = 100 :
6.8 : 50 : 43 : 11 : 7.7 : 19 : 1.4 : 7.0 : 0.54 : 2.2. These are generally
within the normal range for healthy growth of plants (Marschner
1995; Epstein & Bloom 2004) [for the ratios, also see Knecht &
Göransson (2004) and Watanabe et al. (2007)].
These leaf minerals show significant latitudinal trends (in the unit of
kilometres, see supplementary Appendix S1 for full details)
(P < 0.001; Fig. S3a): Nine minerals (N, P, K, Ca, Mg, S, Si, Fe,
and Na) increase from south to north, whereas Mn and Al display an
opposite trend. Similar to the latitudinal patterns, longitudinal (westto-east) gradients also exhibit a decrease from west to east for all the
minerals but Mn (P < 0.001 for all but except Al with P = 0.43)
(Fig. S3b).
Climatic influence on leaf minerals
Temperature and precipitation are the two most critical climatic
variables that shape vegetation distribution and structure (Woodward
1987; Brown & Lomolino 1998), but there are a number of ways they
can be presented, including both the means and variability at annual
and growing season scales (Reich & Oleksyn 2004). Considering the
significant correlations among climatic variables (Table S2), we first
quantify the role of MAT and MAP in shaping the biogeographic
patterns of leaf minerals and then incorporate other climatic variables
into the analysis.
We found that all leaf minerals were significantly correlated with
both MAT and MAP (all P < 0.01; models 1 and 2 in Table S3;
Fig. 2a, b). On average, MAP and MAT explained 10% and 6% of
total variation in the 11 elements, respectively. When both climatic
variables were entered into a stepwise multiple regression (SMR), with
each of the leaf minerals as the dependent variable, MAP played
a significant role for all elements except P and Al, whereas MAT had
no significant influence on Mg, Si, Fe and Mn (i.e. MAT was removed
from SMR; P > 0.10), or explained much less of the variance than
25
1.5
0.6
Mass
Atom
0.4
1.2
0.015
0.010
15
1140
10
0.0
1227
0.005
Al
Fe
Mn
0.000
0.6
0.3
681
5
0
0.9
0.2
678
N
K
Ca
Si
! 2011 Blackwell Publishing Ltd/CNRS
733
819 1475
Mg
S
Na
Leaf mineral
P
678 1161 1060
Al
Fe
Mn
0.0
Atom number (mol kg–1)
Concentration (g kg–1)
1137
20
0.020
Figure 1 The average concentration of 11 leaf minerals (N, P, K,
Ca, Mg, S, Si, Fe, Na, Mn and Al) in China!s plants. Green bars
and blue solid circles are the geometric means on mass and atom
basis, respectively; and the whisker and number on each bar
denote geometric standard error and species number.
Biogeography and variability of leaf chemistry 791
Letter
1.0
1.5
–1.0
1.0
–3.0
All
0.5
2.5
2.5
1.5
1.5
0.5
–0.5
K
–0.5
–1.5
2.5
1.0
1.5
0.0
0.5
–2.0
S
–0.5
–1.5
3.0
2.0
1.0
0.0
–1.0
N
–2.0
1.0
1.5
0.5
–1.0
1.0
–0.5
–1.5
Ca
Si
0.0
–1.0
2.0
1.0
0.0
–1.0
–2.0
–3.0
2.0
1.0
0.0
–1.0
–2.0
–3.0
2.0
1.0
1.5
0.5
–1.0
1.0
–0.5
All
0.5
N
–1.5
2.5
2.0
1.5
1.5
1.0
–0.5
0.5
K
–0.5
2.0
–1.5
3.0
1.0
2.0
0.0
1.0
–1.0
–2.0
S
0.0
–1.0
3.0
1.0
0.0
–2.0
–3.0
Ca
Si
2.0
–1.0
Na
–4.0
–10 0 10 20 30
o
MAT ( C)
Mg
Fe
Al
0 1 2 3 4
MAP (103 mm)
–3.0
All
2.0
1.5
1.0
0.5
0.0
K
2.0
0.0
–2.0
N
S
–1.5
2.0
2.0
1.0
1.0
0.0
–1.0
Ca
2.0
1.0
–1.0
0.5
1.5
0.0
–1.0
0.0
0.0
–1.0
3.0
2.0
1.0
0.0
–1.0
–2.0
Mg
1.0
1.0
–1.0
P
Si
–2.0
Fe
2.0
1.0
0.0
–1.0
Na
Mn
Al
–3.0
–4.0
–2.0
4 6 8 10
4 6 8 10
4 6 8 10
Top soil pH
Top soil pH
Top soil pH
1.5
2.5
0.5
P
1.0
3.0
–3.0
Log10 leaf mineral (mg g–1 )
0.5
–0.5
2.0
Na
Mn
–3.0
–4.0
0 1 2 3 4
0 1 2 3 4
MAP (103 mm)
MAP (103 mm)
(b)
2.0
0.5
2.0
–1.0
(c) 3.0
1.5
Log10 leaf mineral (mg g–1 )
2.0
Log10 leaf mineral (mg g–1)
(a) 3.0
Mn
–10 0 10 20 30
o
MAT ( C)
0.0
–1.0
2.0
1.0
0.0
–1.0
–2.0
–3.0
2.0
1.0
0.0
–1.0
–2.0
–3.0
P
Mg
Fe
Al
–10 0 10 20 30
o
MAT ( C)
MAP for N, K, Ca, S and Na (model 3 in Table S3). Even after adding
other climatic variables (APS, DRT and GSL) into the predictor lists
of the SMR models, MAP explained more of the variance than other
climate variables for most of the leaf minerals (Ca, Mg, S, Si, Fe, Na
and Mn) (Table S4).
Edaphic influence on leaf minerals
Terrestrial plants take up most of their nutrient minerals directly from
soils. Soil chemical attributes (e.g. pH and mineral nutrient availability)
are critical to plant growth and thus affect leaf mineral patterns (Foulds
1993; Vitousek & Farrington 1997; Pärtel 2002; Lynch & St. Clair 2004).
We documented soil mineral and pH values to address their
relationships with leaf mineral concentrations. As often observed
elsewhere, our data demonstrated general positive correlations between
plant leaf and soil mineral contents for most minerals (Table 1; Fig. S2).
Figure 2 Trends in plant leaf chemistry along the climatic and soil pH
gradients in China. (a) Mean annual precipitation (MAP); (b) mean annual
temperature (MAT). (c) Top soil pH. All P < 0.001 for MAP (model 1)
except Al (P = 0.01) and for MAT (model 2) except Na (P = 0.006) in
Table S3 and for soil pH except Si (P = 0.004).
The A or O horizon soil pH value was significantly and positively
correlated with nine of the 11 leaf minerals (N, P, K, Ca, Mg, S, Si, Fe and
Na) and negatively correlated with Mn and Al (all P < 0.005) (Fig. 2c).
Variation in leaf minerals among functional types
Species composition greatly affects the leaf mineral geography (Hou
1982; Reich & Oleksyn 2004) and biogeochemical cycling (Cornwell
et al. 2008). To demonstrate this effect, we divided the seed plants
(the largest part of the dataset) into six functional types: deciduous
broadleaves, deciduous conifers, evergreen broadleaves, evergreen
conifers, grasses and forbs. ANOVA results showed that there were
significant differences in leaf minerals among different plant
functional types (Table S5). In general, forb leaves were richest in
most of the minerals; whereas grass foliage had low mineral
concentrations, except for the highest Si concentration. The
! 2011 Blackwell Publishing Ltd/CNRS
792 W. X. Han et al.
Letter
Table 1 Spearman!s rank correlations (q) between leaf minerals and soil nutrients.
of these mineral elements (all data pooled) was significantly negatively
correlated with both the index of their physiological requirement
(P < 0.005, r2 = 0.77) (Fig. 3a) and the mean concentration of the
elements (P < 0.005, r2 = 0.70) (Fig. 3b). Both the CV and the mean
concentration of the elements were also significantly correlated with
the rank order index of relative limitation (Spearman!s q = 0.95,
P < 0.001 and q = )0.76, P = 0.028, respectively).
The 11 elements differ in their sensitivities to the gradients in
climate and soil (Figs. 2 and 4). Na, Mn, Si and Al respond to climate
(MAP ⁄ MAT) with steeper RMA slopes than the other elements, with
N (followed by P and K) showing the shallowest responses to MAP
and MAT (Fig. 4). The sensitivity of concentrations (the absolute
value of the regression slopes) to MAP and MAT were positively
correlated with the index of relative limitation (Spearman!s q = 0.88,
P < 0.005 for both MAP and MAT). There was also a positive
correlation (Spearman!s q = 0.83, P = 0.010) between the RMA
slopes of mineral concentrations vs. soil pH and the relative limitation
rank.
The above results, together with an allometric relation between the
mean concentrations of the minerals and their physiological requirements (Fig. 3c), suggest that nutrients required in a high concentration
in leaves and considered most frequently limiting in environment
should show a small variation in their concentration and lower
sensitivity to the environmental factors (Stability of Limiting Elements
Hypothesis), which is further discussed below.
Leaf minerals (mg g)1) are log10-transformed before analysis. Soil total N is density
based (kg m)2), and the other soil nutrients are content based (mg g)1), which are
divided into four levels, respectively (also see Figure S2 for details)
q
N
P
K
Ca
Mg
S
Si
Fe
Na
Mn
Al
n
0.064
0.350
0.272
0.223
0.347
0.199
)0.077
)0.036
0.281
0.433
0.179
P
1910
2510
2009
2052
932
528
800
1864
1320
1584
1086
0.005
0.000
0.001
0.001
0.001
0.001
0.030
0.119
< 0.001
< 0.001
< 0.001
<
<
<
<
<
deciduous angiosperms (with short leaf life-span) had
mineral-rich (except for Mn and Al) leaves in contrast
evergreen ones; while conifers were lower in most of
minerals, compared with the broadleaved counterparts (see
for details).
generally
with the
the leaf
Table S5
Differences in stability and sensitivity of limiting elements
The relative variability (or stability) of the 11 mineral concentrations
can be demonstrated by their coefficient of variation (CV, %): the CVs
increase from 41, 67, 77, 81, 94, 126, 159, 175, 188, 266 to 479 for N,
P, K, Mg, Ca, Si, Fe, S, Al, Na and Mn, respectively (Table S1).
A similar pattern of CV still holds for the functional-type averaged
concentration of each element: from 29, 36, 38, 39, 44, 53, 60, 60, 63,
85 to 101 (%) for N, P, K, Mg, Fe, Ca, Al, Si, S, Mn and Na, despite
the small sample size (n = 7, fern and six seed-plant groups). The CV
500
Mn
400
300
200
100
S
Fe
Ca
Mg
K
P
0
–1.5
Biogeographic patterns and the environmental control
Leaf minerals showed large variations among China!s terrestrial plants
(Fig. 1; Table S1) and exhibited significant latitudinal and longitudinal
trends for each element (Fig. S3). If we translate latitudinal and
(b) 600
CV of leaf element (%)
CV of leaf element (%)
(a) 600
DISCUSSION
–1.0
–0.5
0
N
0.5
1.0
1.5
Log10 elemental requirement (mg g–1)
500 Mn
400
300
200
100
0
–1.0
Na
Al
S
Fe
P
–0.5
0
Si
Mg
0.5
Ca
K
1.0
N
1.5
Log10 leaf element (mg g–1)
Log10 leaf element (mg g–1)
(c) 1.5
N
Ca
1.0
K
Mg
0.5
S
0
P
–0.5
Fe
–1.0
Mn
–1.5
–1.5
–1.0
–0.5
0
0.5
1.0
1.5
Log10 elemental requirement (mg g–1)
! 2011 Blackwell Publishing Ltd/CNRS
Figure 3 Relationships between plant elemental requirement,
mean leaf elemental concentration and its coefficient of variation
(CV). (a) Elemental requirement vs. CV, log10(y) = 2.067–
0.407 * x (r2 = 0.77, P < 0.005); (b) leaf elemental concentration
vs. CV, log10(y) = 2.212–0.423 * x (r2 = 0.70, P < 0.005); (c)
elemental requirement vs. leaf elemental concentration,
y = 0.182 + 0.859 * x (r2 = 0.96, P < 0.005). 95% confidence
bands for all the fitting curves are also shown.
Biogeography and variability of leaf chemistry 793
Letter
40
30
Total
20
50
MAT
MAP
Soil
40
20
10
10
0
RMA regression slope (%)
–10
N
P
K Ca Mg S
Si Fe Na Mn Al
0
–10
–20
–20
–30
–30
–40
–40
40
40
30
P
K Ca Mg S
Si Fe Na Mn Al
Evergreen broadleaf
20
10
10
0
N
P
K Ca Mg S
Si Fe Na Mn Al
0
–10
–20
–30
–20
–40
–30
40
40
30
N
30
Forb
20
20
10
10
P
K Ca Mg S
Si Fe Na Mn Al
Grass
0
0
–10
N
30
Deciduous broadleaf
20
–10
Evergreen conifer
30
N
P
K Ca Mg S
Si Fe Na Mn Al
–10
N
P
K Ca Mg S
Si Fe Na Mn Al
–20
–20
Leaf mineral element
Figure 4 Reduced Major Axis (RMA) regression slopes for leaf chemistry against MAT ⁄ MAP ⁄ soil chemistry for five functional types. RMA slopes for the three variables
(MAT, MAP and Soil) were transformed to eliminate effects of the different units (see the Methods). Deciduous conifers and ferns are not shown because of their small sample
sizes. The segmental lengths of the bars represent directly the slopes of the regression lines between leaf chemistry and temperature, precipitation and soil chemistry. Bars rather
than points are shown here because the bars provide a clearer visual picture of the changes in the slopes across the different functional types and environmental conditions.
Both leaf chemistry and MAP are in log-scale.
longitudinal degrees into a spatial distance basis, the rates of the
change of these leaf minerals displayed a surprising close consistency
in both directions (r2 = 0.94, P < 0.001; Fig. 5). The similar rates of
the change of leaf mineral concentrations along the two gradients
suggest similar underlying biophysical (environmental) and biological
controls that shape the biogeographic patterns of the plant minerals.
Climatically, the north-to-south and west-to-east gradients in China
both reflect shifts from cold, dry to warmer, moister conditions,
although the thermal gradient is steeper in the former and the
moisture gradient more pronounced in the latter (Fig. S4). Additionally, both the latitudinal and longitudinal gradients in plant mineral
concentrations are associated with, and likely reflect, pervasive
geographic patterns in the structure and function of terrestrial
ecosystems (such as functional type, biodiversity, soil development,
vegetation primary production and ecological traits of plants) (Brown
& Lomolino 1998; Hedin 2004; Reich & Oleksyn 2004; Wright et al.
2004), which themselves reflect responses to climate gradations.
Climate, soil nutrient contents and species composition all influence
plant mineral biogeography in complex ways (Hou 1982; Reich &
Oleksyn 2004). General linear models (GLM) involving climate, soil
and plant functional types (indicate that these three factors together
account for a substantial part of the biogeographic variation in the
concentrations of these leaf minerals (full models in Table S6): 37%
for leaf N, and more than 20% for the other minerals except Fe and S
(both less than 16%). In addition, the explanatory power of these
three factors for different minerals varied greatly. Plant functionaltype variation accounted for the largest explained fraction of the
variances for leaf N, P, K, Ca, Mg, Si, Fe, Mn and Al, while climate
explained the most for S and Na (Table S6).
However, significant collinearities between these factors could
potentially obscure their true roles. Partial GLM regressions (Heikkinen et al. 2005) can separate the variance explained by multiple factors
into independent effects of all individual factors and their interactive
effects with the remaining factors (Legendre & Legendre 1998).
Performing the partial GLM indicated that the independent effects of
functional type were much larger than those of climate and soil nutrient
contents, for leaf N, P, K, Ca, Mg, Si and Mn (Table S6). By contrast,
the independent effects of climate on S, Na and Al were the largest.
! 2011 Blackwell Publishing Ltd/CNRS
794 W. X. Han et al.
Letter
Mean rate ( g g–1 km–1), longitude
4
y = 1.004x – 0.012
3
r 2 = 0.996
Ca
2
1
y = 0.880x + 0.042
r 2 = 0.943
0
–1
–1
0
1
2
3
Mean rate ( g g–1 km–1), latitude
4
Figure 5 Relationship between the mean rates of change of the 11 leaf minerals in
China!s plants along latitudinal and longitudinal gradients on spatial distance basis.
The close correlation (r2 = 0.943 and 0.996 with and without Ca, respectively;
P < 0.001) indicates almost identical mean rates of change along these two
gradients. The blue and green lines denote the regressions of longitudinal mean rate
(y) vs. the latitudinal mean rate (x), with and without Ca, respectively. Note that the
latitudinal gradient is expressed as distance (km) from the equator at certain
longitude and the longitudinal gradient as distance (km) from the prime meridian at
certain latitude.
These results suggest that the geography of plant leaf minerals was
largely controlled by plant functional types, favouring the species
composition hypothesis (Reich & Oleksyn 2004). This functional-type
effect can also be illustrated by the generally consistent (parallel)
relationships between the deciduous percentage and the leaf mineral
contents (except for Mn and Al) in the woody plants (deciduous vs.
evergreen broadleaf) along the latitudinal gradient (Fig. S5). Species
that tend to have high mineral concentrations, such as deciduous
(relative to evergreen) plants and herbaceous (relative to woody)
plants (Table S5), are more proportionately distributed in northern
than southern regions, and in dry west inland than humid coastal areas
(Hou 1983; Gaston 2000).
In previous studies, the effects on leaf traits of environmental
factors and plant functional types and their interactions have rarely
been analysed (but see Reich et al. 2007). In the current study, the
interactive effects of climate and functional type accounted for 7.5%
of the variation in leaf N, and those of climate, functional type and soil
accounted for another 9.3% (Table S6; Fig. S6). The interaction of
functional type, climate and soil accounted for significant portions of
the variation in leaf P and Mn (8.6% and 10.5%, respectively). Note
that the interactive effects on certain minerals will be negative when
the relationship between two factors is mainly suppressive rather than
additive (e.g. the interactive effects of climate and functional type on
leaf Al) (Chevan & Sutherland 1991; Heikkinen et al. 2005).
Variation and sensitivity in relation to physiological requirements
and relative limitation
We posit that both physiology and ecological stoichiometry should
result in variability being increasingly constrained as elemental relative
supply limitations grow and requirements increase (Ingestad 1997;
! 2011 Blackwell Publishing Ltd/CNRS
Sterner & Elser 2002). However, heretofore it was unknown whether
variability per se would differ among elements in any relation to their
ecology. Equally unknown was whether different mineral elements
vary in relation to environmental gradients in idiosyncratic or
patterned fashion. We hypothesised here that elements with
high physiological requirements, high average concentrations and
most frequently limiting in nature would be more stable and less
sensitive to environmental gradients (the Stability of Limiting
Elements Hypothesis).
The stability of limiting elements can be explained as follows. For
elements with high requirement and that are frequently limiting,
extreme low values should be limited by stoichiometric requirements
(i.e. growth would be suboptimal with concentrations below a certain
threshold). Extreme high values are also less likely because higher
supply would often lead to higher growth rate and thus elemental
dilution (Marschner 1995).
The mean concentrations of the minerals were well correlated with
an index of their physiological requirements (Fig. 3c) (log–log
scaling, P < 0.001, r2 = 0.96). The RMA slope (0.87) of this log–
log relationship is < 1 (although for this difference, P = 0.138,
indicating only 86% likelihood that this is not by chance). A log–log
slope < 1 is consistent with the idea that the extent to which the
minimum requirement is exceeded would be less in elements required
at the greatest concentrations. Additionally, P is the only element
outside of the 95% confidence intervals of the relationship, and also
has a mean concentration less than the index of physiological
requirement, consistent with the idea that it is often found at limiting
levels.
The patterns of CV of leaf minerals are consistent with the Stability
of Limiting Elements Hypothesis. N, P and K (the three most
frequently limiting nutrients) had the lowest CV values (41, 67 and 77,
respectively), while Al, Na and Mn (the most often toxic elements) the
highest (188, 266 and 479, respectively) (Table S1). In addition, the
CV of these mineral elements (which was negatively correlated to both
the index of physiological requirement and the mean concentration;
see Fig. 3a,b) showed significantly positive correlation with the rank
order index of relative limitation, suggesting that the more frequently
limiting in nature, the more stable the concentration of an element
should be. It is possible that geomorphological sources of variation
could vary widely among elements (Hou 1982), leading to variation in
their CV which would be unrelated to their requirements or rank
order of limitation. However, the strength of the observed relationships (Fig. 3) suggests that such geomorphological "noise! is low
compared with the "signal! from physiological requirement and
stoichiometry. For example, there is no significant correlation between
the index of relative limitation and the mean contents of the soil
elements or their CVs (Spearman!s q = )0.095, P = 0.82 and
q = )0.50, P = 0.21, respectively).
Results from the regression analyses were also consistent with the
prediction by the Stability of Limiting Elements Hypothesis.
According to the hypothesis, the environmental sensitivity (assessed
as the RMA regression slope) should be lowest for elements that are
required in the highest concentrations, most abundant, and most
often limiting in nature. Leaf N and P display shallower slopes
against the climate and soil gradients, compared with leaf Mn and
Na (Figs. 2 and 4); and there exist positive correlations between the
index of relative limitation and the absolute value of the slopes
against MAP ⁄ MAT or soil gradients. These positive correlations
imply that mineral elements considered more frequently limiting tend
Biogeography and variability of leaf chemistry 795
Letter
to vary less across the climate (e.g. precipitation) or soil (e.g. pH)
gradients.
Implications of mineral variation across functional types
The RMA analyses indicated some similarity of overall functional type
responses to MAP, MAT and soil nutrient concentration, but also
considerable variation (Fig. 4). The analysis by functional type
indicates that grasses and forbs respond in a similar fashion to
MAP, MAT and soil nutrients. However, the tree functional types
differ in all of these aspects, with evergreen broadleaf trees the most
divergent. Therefore, if gradient analyses are any indication, future
changes in climate will exert idiosyncratic effects on plant functional
types, as well as modifying nutrient biogeography by altering spatial
patterns of composition. Recent experimental imposition of warming,
and reduced precipitation, on Mediterranean vegetation (Peñuelas
et al. 2008; Sardans et al. 2009) led to species-specific differences in the
response of a range of plant macro- and micro-nutrients. Those
species that showed the greatest change in nutrients were also most
affected in terms of growth.
CONCLUSION
Our study is, to our knowledge, the first to comprehensively
document the foliar chemistry of multiple mineral elements and
quantify the potential controls and variability at a large scale. Of the
three major factors considered to influence the biogeographic
distribution, functional type shows the greatest direct influence for
most leaf minerals. However, climate and soil are both directly and
indirectly influential, as they contribute substantially to shaping the
distribution of vegetation (species composition) (Brown & Lomolino
1998; Chapin et al. 2002) (Fig. S7). In addition, we found that
variation in elemental concentrations was more constrained (more
stable and less sensitive to the environment) for nutrients with highest
requirements, generally present in the highest concentrations and
considered the generally most limiting to plant growth (the Stability of
Limiting Elements Hypothesis).
Our findings broaden the knowledge of the biogeochemical cycling
of elements through plants and the fundamental constraints on plant
stoichiometry across wide gradients of environmental factors. They
also provide a beginning of synthetic data compilation and analyses
that will eventually make it possible to better parameterise complex
multi-element biogeochemical models, that should be developed in
the future (Hedin 2004; Wright et al. 2004; Lambers et al. 2008).
As nutrients with highest requirements and most limiting in nature
also are globally concentrated in shallower soil horizons (Jobbágy &
Jackson 2001), it appears that physiological requirements in conjunction with biogeochemical availability influence elemental distribution
spatially in soils and biogeographically across climate gradients, as well
as constraining general levels of variability. Other signatures of relative
requirements and availability of elements likely remain to be
discovered at local to global scales.
ACKNOWLEDGMENTS
We thank Y. Zhang, Y.H. Chen, J. S. He, L. Y. Tang and L. P. Li from
Peking University, M. L. Huang, A. H. Tang and X. J. Liu from China
Agriculture University, X. L. Di and W. Lin from Beijing Forestry
University, J. P. Cui from Beijing Botanical Garden and Y. D. Li from
Hainan Station of Chinese Academy of Forest Sciences, for assisting
data collection and field sampling and measurement. The authors also
thank P. Vitousek at Stanford University and the anonymous referees
for their valuable comments on the manuscript. The project was
funded by National Natural Science Foundation of China (Project
Nos. 40973054, 31021001, 30821003), the National Basic Research
Program of China on Global Change (2010CB50600), the U.S.
National Science Foundation LTER Program (DEB-0080382) and the
University of Minnesota Institute on the Environment.
REFERENCES
Aerts, R. & Chapin, F.S. III (2000). The mineral nutrition of wild plants revisited: a
re-evaluation of processes and patterns. Adv. Ecol. Res., 30, 1–67.
Ågren, G.I. (2008). Stoichiometry and nutrition of plant growth in natural communities. Ann. Rev. Ecol. Evol. Syst., 39, 153–170.
Barron, A.R., Wurzburger, N., Bellenger, J.P., Wright, S.J. & Hedin, L.O. (2009).
Molybdenum limitation of asymbiotic nitrogen fixation in tropical forest soils.
Nat. Geosci., 2, 42–45.
Brown, J.H. & Lomolino, M.V. (1998). Biogeography. Sinauer Associates, Sunderland,
MA.
Chapin, F.S. III, Matson, P.A. & Mooney, H.A. (2002). Principles of Terrestrial Ecosystem Ecology. Springer-Verlag, New York, NY.
Chevan, A. & Sutherland, M. (1991). Hierarchical partitioning. The American Statistician, 45, 90–96.
Cornwell, W.K., Cornelissen, J.H.C., Amatangelo, K., Dorrepaal, E., Eviner,
V.T., Godoy, O. et al. (2008). Plant species traits are the predominant control
on litter decomposition rates within biomes worldwide. Ecol. Lett., 11, 1065–
1071.
Elser, J.J., Borer, E.T., Bracken, M.E., Cleland, E.E., Gruner, D.S., Harpole, W.S.
et al. (2007). Global analysis of nitrogen and phosphorus limitation of primary
producers in freshwater, marine and terrestrial ecosystems. Ecol. Lett., 10, 1135–
1142.
Epstein, E. & Bloom, A.J. (2004). Mineral Nutrition of Plants: Principles and Perspectives.
2nd edn Sinauer Associates, Sunderland, MA.
Fang, J.Y., Song, Y.C., Liu, H.Y. & Piao, S.L. (2002). Vegetation-climate relationship and its application in the division of vegetation zone in China. Acta Bot. Sin.,
44, 1105–1122.
Foulds, W. (1993). Nutrient concentrations of foliage and soil in south-western
Australia. New Phytol., 125, 529–546.
Gaston, K.J. (2000). Global patterns in biodiversity. Nature, 405, 220–227.
Han, W.X., Fang, J.Y., Guo, D.L. & Zhang, Y. (2005). Leaf nitrogen and phosphorus stoichiometry across 753 terrestrial plant species in China. New Phytol.,
168, 377–385.
Hedin, L.O. (2004). Global organization of terrestrial plant-nutrient interactions.
Proc. Nat. Acad. Sci. USA, 101, 10849–10850.
Heikkinen, R.K., Luoto, M., Kuussaari, M. & Poyry, J. (2005). New insights into
butterfly-environment relationships using partitioning methods. Proc. Royal Soc.
(B), 272, 2203–2210.
Hou, H.Y. (1982). The Vegetation Geography of China with Reference to the Chemical
Composition of Dominant Plants. Science Press, Beijing.
Hou, H.Y. (1983). Vegetation of China with reference to its geographical distribution. Ann. Mo. Bot. Gard., 70, 509–549.
Hungate, B.A., Stiling, P.D., Dijkstra, P., Johnson, D.W., Ketterer, M.E., Hymus,
G.J. et al. (2004). CO2 elicits long-term decline in nitrogen fixation. Science, 304,
1291.
Ingestad, T. (1997). A shift of paradigm is needed in plant science. Physiol. Plant.,
101, 446–450.
Jobbágy, E.G. & Jackson, R.B. (2001). The distribution of soil nutrients with depth:
global patterns and the imprint of plants. Biogeochemistry, 53, 51–77.
Kaspari, M., Garcia, M.N., Harms, K.E., Santana, M., Wright, S.J. & Yavitt, J.B.
(2008). Multiple nutrients limit litterfall and decomposition in a tropical forest.
Ecol. Lett., 11, 35–43.
Knecht, M. & Göransson, A. (2004). Terrestrial plants require nutrients in similar
proportions. Tree Physiol., 24, 447–460.
! 2011 Blackwell Publishing Ltd/CNRS
796 W. X. Han et al.
Lambers, H., Raven, J.A., Shaver, G.R. & Smith, S.E. (2008). Plant nutrientacquisition strategies change with soil age. Trends Ecol. Evol., 23, 95–103.
Legendre, P. & Legendre, L. (1998). Numerical Ecology, 2nd edn. Elsevier Science
BV, Amsterdam.
Lynch, J.P. & St. Clair, S.B. (2004). Mineral stress: the missing link in understanding
how global climate change will affect plants in real world soils. Field Crops Res.,
90, 101–115.
Marschner, H. (1995). Mineral Nutrition of Higher Plants. Academic Press, New York,
NY.
Marschner, P. & Rengel, Z. (2007). Nutrient Cycling in Terrestrial Ecosystems. SpringerVerlag, Heidelberg.
McGroddy, M.E., Daufresne, T. & Hedin, L.O. (2004). Scaling of C:N:P stoichiometry in forests worldwide: implications of terrestrial redfield-type ratios.
Ecology, 85, 2390–2401.
Pärtel, M. (2002). Local plant diversity patterns and evolutionary history at the
regional scale. Ecology, 83, 2361–2366.
Peñuelas, J., Sardans, J., Ogaya, R. & Estiarte, M. (2008). Nutrient stoichiometric
relations and biogeochemical niche in coexisting plant species: effect of simulated
climatic change. Polish J. Ecol., 56, 613–622.
R Development Core Team. (2005). R: a language and environment for statistical
computing. R Foundation for Statistical Computing, Vienna, Austria. Available
at: http://www.Rproject.org/bin/windows/base/old/2.2.1/. Last accessed
1 November 2005.
Reich, B.P. (2005). Global biogeography of plant chemistry: filling in the blanks.
New Phytol., 168, 263–266.
Reich, P.B. & Oleksyn, J. (2004). Global patterns of plant leaf N and P in relation to
temperature and latitude. Proc. Nat. Acad. Sci. USA, 101, 11001–11006.
Reich, P.B., Oleksyn, J., Modrzynski, J., Mrozinski, P., Hobbie, S.E., Eissenstat,
D.M. et al. (2005). Linking litter calcium, earthworms and soil properties: a
common garden test with 14 tree species. Ecol. Lett., 8, 811–818.
Reich, P.B., Wright, I.J. & Lusk, C.H. (2007). Predicting leaf physiology from simple
plant and climate attributes: a global GLOPNET analysis. Ecol. Appl., 17, 1982–
1988.
Sardans, J., Peñuelas, J., Prieto, P. & Estiarte, M. (2009). Changes in Ca, Fe, Mg,
Mo, Na, and S content in a Mediterranean shrubland under warming and
drought. J. Geophys. Res., 113, G03039.
Sokal, R.R. & Rohlf, F.J. (1995). Biometry: The Principles and Practice of Statistics in
Biological Research, 3rd edn Freeman, New York, NY.
Sterner, R.W. & Elser, J.J. (2002). Ecological Stoichiometry: The Biology of Elements from
Molecules to the Biosphere. Princeton University Press, Princeton, NJ.
Townsend, A.R., Cleveland, C.C., Houlton, B.Z., Alden, C.B. & White, J.W.C.
(2011). Multi-element regulation of the tropical forest carbon cycle. Front. Ecol.
Environ., 9, 9–17.
Van Groenigen, K.J., Six, J., Hungate, B.A., DeGraaff, M.A., Van Breemen, N. &
Van Kessel, C. (2006). Element interactions limit soil carbon storage. Proc. Nat.
Acad. Sci. USA, 103, 6571–6574.
Vitousek, P.M. & Farrington, H. (1997). Nutrient limitation and soil development:
experimental test of a biogeochemical theory. Biogeochemistry, 37, 63–75.
Vitousek, P.M., Porder, S., Houlton, B.Z. & Chadwick, O.A. (2010). Terrestrial
phosphorus limitation: mechanisms, implications, and nitrogen-phosphorus
interactions. Ecol. Appl., 20, 5–15.
Watanabe, T., Broadley, M.R., Jansen, S., White, P.J., Takada, J., Satake, K. et al.
(2007). Evolutionary control of leaf element composition in plants. New Phytol.,
174, 516–523.
White, P.J. & Brown, P.H. (2010). Plant nutrition for sustainable development and
global health. Ann. Bot., 105, 1073–1080.
! 2011 Blackwell Publishing Ltd/CNRS
Letter
Woodward, F.I. (1987). Climate and Plant Distribution. Cambridge University Press,
Cambridge.
Wright, I.J., Reich, P.B., Westoby, M., Ackerly, D.D., Baruch, Z., Bongers, F. et al.
(2004). The worldwide leaf economics spectrum. Nature, 428, 821–827.
Zhang, X.S. (2007). Atlas of China!s Vegetation (1:100 M). Geological Publishing
House, Beijing.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the online
version of this article:
Appendix S1 Full description of the Material and Methods.
Data S1 Dataset on concentrations of 11 leaf minerals in terrestrial
plants of China and associated information.
Figure S1 Distribution of the samples for 11 leaf minerals in plants of
China.
Figure S2 Relationships between minerals in China!s soils and plant
leaves.
Figure S3 Geographic patterns of plant leaf minerals in China.
Figure S4 Geographic patterns of temperature and precipitation in the
mainland and two large islands of China.
Figure S5 Latitudinal trends in leaf minerals of deciduous vs.
evergreen broadleaves.
Figure S6 Variation partitioning of environmental factors in accounting for the variations in leaf nitrogen and phosphorus concentrations.
Figure S7 The schematic diagram showing the climatic controls on the
leaf mineral patterns.
Table S1 Leaf mineral concentrations in 1900 plant species in China.
Table S2 Correlations of five climatic variables, using climate data in
all sites in this study.
Table S3 Linear regressions of leaf minerals on MAT and MAP.
Table S4 Model summary for the stepwise multiple regression of leaf
minerals on five climatic variables.
Table S5 Leaf minerals for different functional types of seed plants in
China.
Table S6 Summary of the (partial) general linear models for the effects
of climate, plant functional type and soil nutrient contents on leaf
minerals.
As a service to our authors and readers, this journal provides
supporting information supplied by the authors. Such materials are
peer-reviewed and may be re-organised for online delivery, but are not
copy edited or typeset. Technical support issues arising from
supporting information (other than missing files) should be addressed
to the authors.
Editor: Richard Lindroth
Manuscript received 22 February 2011
First decision made 1 April 2011
Manuscript accepted 16 May 2011